Table 3.

Open syncmer versus minimizer mappings for two versions of minimap2 on long-reads. We fix parameters $(k,s,t,w)=(15,11,3,9)$ so the density is 1/5 for both seeding methods

Long-read dataset	Reference	$|OS\cap \overline{M}|$ (mapQ)	$|M\cap \overline{OS}|$ (mapQ)	$\left|OS\right|$	$\left|M\right|$	Total no. of reads	% increase in mapped reads
E.coli W (bc1087)	E.coli W (bc1087)	312 (21.56)	102 (11.57)	194 455	194 245	196 901	0.108
E.coli K12 (bc1106)	E.coli W (bc1087)	548 (20.30)	187 (11.33)	220 459	220 098	226 906	0.164
K.pneumoniae (bc1074)	E.coli W (bc1087)	11 434 (19.80)	3679 (12.10)	143 724	135 969	251 838	5.70
Downsampled human ONT (rel3)	Human—CHM13	370 (3.53)	103 (2.52)	37 819	37 552	51 210	0.711
Downsampled human ONT (rel3)	Mouse—GRCm38	2467 (2.32)	1005 (1.90)	19 214	17 752	51 210	8.23

Note: OS is the subset of reads successfully mapped using open syncmers, and M similarly for minimizers. $OS\cap \overline{M}$ is the set of reads which are uniquely mapped by open syncmers, and $M\cap \overline{OS}$ are reads uniquely mapped by minimizers. The average mapQ outputted by minimap2 within the set is presented as well (Section 5.2).